Download Modelling the impact of future changes in climate, CO2

ARTICLE IN PRESS
Global Environmental Change 14 (2004) 21–30
Modelling the impact of future changes in climate, CO2 concentration
and land use on natural ecosystems and the terrestrial carbon sink
P.E. Levy*, M.G.R. Cannell, A.D. Friend
Centre for Ecology and Hydrology, Bush Estate, Penicuik, Midlothian EH26 0QB, UK
Abstract
We used a global vegetation model, ‘HyLand’, to simulate the effects of changes in climate, CO2 concentration and land use on
natural ecosystems. Changes were prescribed by four SRES scenarios: A1f, A2, B1 and B2. Under all SRES scenarios simulated, the
terrestrial biosphere is predicted to be a net sink for carbon over practically all of the 21st century. This sink peaks around 2050 and
then diminishes rapidly towards the end of the century as a result of climate change. Averaged over the period 1990–2100, the
net sink varies between scenarios, from B2 to 6 Pg C yr 1. Differences are largely the result of differences in CO2 concentrations.
Effects of climate change are substantially less, and counteract the effect of elevated CO2. Land use change results in a loss of
carbon to the atmosphere in the B2B scenario, in which the increase in cropland area continues. In the other scenarios, there is a
decrease in croplands and grassland, with a corresponding increase in natural vegetation, resulting in a net sink to the biosphere. The
credibility of these results depends on the accuracy of the predictions of land use change in the SRES scenarios, and these are highly
uncertain.
As CO2 is the dominating influence on the vegetation, the scenarios with high fossil fuel emissions, and thus the highest CO2
concentrations (A1F & A2) generate the largest net terrestrial sink for carbon. This conclusion would change if these scenarios
assumed continued deforestation and cropland expansion. Without the beneficial effects of elevated CO2, the effects of climate
change are much more severe. This is of concern, as the long-term and large-scale effects of elevated CO2 are still open to question.
Differences between scenarios in the predicted global spatial pattern of net biome productivity and vegetation type are relatively
small, and there are not major shifts in the dominant types. The regions predicted to be at greatest risk from global environmental
change are Amazonia, the Sahel, South Central USA and Central Australia.
r 2003 Elsevier Ltd. All rights reserved.
1. Introduction
Global climate change will have impacts upon natural
vegetation, affecting ecological and physiological processes, altering growing season length, biomass production, competition, and leading to shifts in species ranges
and possible extinctions. As well as affecting climate,
increases in the global atmospheric CO2 concentration
will affect natural vegetation directly. Many experiments
have shown that increased CO2 concentrations can
increase photosynthesis and growth, and reduce stomatal conductance, though the response is variable between
species and experimental conditions (Jarvis, 1998). The
main human-induced factors changing vegetation in the
recent past have been land use changes such as logging
*Corresponding author. Tel.: +44-131-445-4343; fax: +44-131-4453943.
E-mail address: [email protected] (P.E. Levy).
0959-3780/$ - see front matter r 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.gloenvcha.2003.10.005
for timber and clearance for agricultural use (Houghton,
1995). With the global population expected to nearly
double over the next century, this is expected to
continue to be a major influence.
Terrestrial ecosystems represent a large store of
carbon, approximately three times that of the atmosphere. Changes in terrestrial ecosystems induced by any
of the above causes are likely to result in changes in the
amount of biomass and/or soil organic matter, and will
result in a flux of carbon between the land and the
atmosphere. Positive [or negative] feedback loops are
possible if this carbon flux reinforces [or counteracts] the
original change in the climate system (Cox et al., 2000).
Several models have been used to assess the impact of
future climate change on global vegetation. These tend
to be empirical in their treatment of plant physiology,
e.g. several do not include CO2 concentration as a
variable in the calculation of primary production
(Cramer et al., 1999). Very few incorporate land use
ARTICLE IN PRESS
P.E. Levy et al. / Global Environmental Change 14 (2004) 21–30
22
change, so model output represents potential vegetation,
rather than the vegetation that is actually present (see
McGuire et al., 2001; DeFries et al., 1999). Few studies
have assessed the integrated effects of changes in
climate, CO2 concentration and land use on global
vegetation over the next century. Here, we used the
HyLand model to simulate the future impact of all of
these global change factors, as prescribed by the IPCC
Special Report on Emission Scenarios (SRES), on the
functioning and distribution of ecosystems.
Specifically, we aimed to:
(i) Quantify the carbon source/sink strength of the
terrestrial biosphere.
(ii) Quantify the contribution of the various factors to
the net carbon flux.
(iii) Predict the future global distribution of vegetation
types.
(iv) Identify regions under greatest threat from global
change.
2. Methods
2.1. Model
A dynamic global vegetation model (Hybrid) was
developed previously to predict the impact of future
climates on global vegetation (Friend et al., 1997; Friend
and White, 2000). The model is process-based and is
able to simulate transient changes in the growth and
distribution of different biomes year by year. A further
version of the model (HyLand) was constructed to
simulate the transient effects of changes in land use on
vegetation properties and carbon stocks. Changes to the
model of Friend et al. (1997) are described in Table 1
and Appendix A.
2.2. SRES scenarios
The future emissions of greenhouse gases will depend
on future trends in the global economy, population
growth, and technological change. The recognition that
these factors are linked led the IPCC to develop a set of
scenarios, in which change in each factor is consistent
with a particular ‘storyline’, published as the SRES
(IPCC, 2000). Four storylines were defined to encompass a range of possible ways in which socio-economics
may develop over the next century. The storylines are
differentiated in two main dimensions: the balance
between economic and environmental concerns, and
the heterogeneity of regional development patterns. The
four storylines can be briefly characterised as follows
(see Arnell et al., 2004).
The A1 scenario envisages very rapid economic
growth with increasing globalisation, an increase in
general wealth, with convergence between regions and
reduced differences in regional per capita income.
Materialist–consumerist values predominant, with rapid
technological change and low population growth. Three
variants within this family make different assumptions
about sources of energy for this rapid growth: fossil
intensive (A1F), non-fossil fuels (A1T) or a balance
across all sources (A1B).
The A2 scenario is a heterogeneous, market-led
world, with less rapid population and economic growth
than A1. The underlying theme is self-reliance and
preservation of local identities. Economic growth is
regionally orientated, and hence both income growth
and technological change are regionally diverse. Fertility
patterns across regions converge slowly, resulting in
high population growth.
The B1 scenario has the same low population growth
as A1, but development takes a much more environmentally sustainable pathway, with global-scale
Table 1
Processes represented in HyLand, the approach taken and the principal source describing it
Process
Competition
Photosynthesis
Plant respiration
Stomatal conductance
Transpiration
Nitrogen uptake
Phenology
Litter production
Allocation
Tree mass/height
Decomposition
Hydrology
Land use change
a
Approach
Source
a
Simple, empirical
Leaf: biochemistry
Canopy: optimisation
Fixed fraction of gross photosynthesisa
Empirical
Energy balance/Penman-monteith equation
Prescribed concentrationsa
Not representeda
Fixed turnover rates
Fixed coefficientsa
Not representeda
Adapted century model
Single layer bucket
Prescribed effects of clear-felling and croppinga
Denotes differences from Friend et al. (1997).
This paper
Farquhar and von Caemmerer (1982)
Sellers et al. (1992)
Waring et al. (1998)
Jarvis (1976), and Stewart (1988)
Friend (1995), and Monteith and Unsworth (1990)
Friend et al. (1997); this paper
This paper
Comins and McMurtrie (1993)
Friend et al. (1997)
This paper
ARTICLE IN PRESS
P.E. Levy et al. / Global Environmental Change 14 (2004) 21–30
800
Cropland (% global land area)
CO2 concentration (ppm)
1000
Historical
A1F
A2
B1
B2
600
16
14
B2
12
10
8
B1
6
A1
4
2
0
1700
400
23
1800
1900
2000
2100
Fig. 2. Global cropland area from historical sources (Ramankutty and
Foley, 1999) and SRES land use change scenarios.
200
1850
1900
1950
2000
2050
2100
Year
Fig. 1. CO2 concentration from historical sources and the four SRES
scenarios.
cooperation and regulation. Clean and efficient technologies are introduced. The emphasis is on global
solutions to achieving economic, social and environmental sustainability.
In the B2 scenario, population increases at a lower
rate than A2, with development following environmentally, economically and socially sustainable, locally
orientated pathways.
The SRES scenarios divide the globe up into four
regions which are considered to be homogeneous socioeconomically: OECD countries (Europe, North America
& Australia); former states of the Soviet Union; Asia;
and Africa & Latin America.
2.3. CO2 concentration data
The conversions between time courses of CO2
emissions and CO2 concentration for each of the four
SRES scenarios were obtained from the ISAM model
(IPCC, 2001). The highest CO2 concentration is
achieved in the A1F scenario, reaching 970 ppm by
2100 (Fig. 1). The B1 scenario has the lowest CO2
concentration, reaching 549 ppm in 2100. The A2 and
B2 scenarios are intermediate.
2.4. Climate data
Greenhouse gas emissions from the four SRES
scenarios were used as inputs to HadCM3, the most
recent version of the Hadley Centre global climate
model. Compared with earlier versions, the model
includes improved representations of some key processes, such as the influence of sulphate aerosols, and
does not require a flux correction to maintain a stable
climate. HadCM3 uses a 96 73 global grid (3.75
longitude 2.5 latitude).
One simulation was performed with HadCM3 for
each of the A1F and B1 scenarios. For the A2 scenario,
three ensemble simulations with the same forcing but
different initial boundary conditions were performed, to
obtain a measure of the natural climatic variability.
Similarly, two ensemble simulations were performed for
the B2 scenarios. Because of the computation time of the
HyLand model only one of these simulations was used
for each scenario, referred to as A2c and B2b.
The HadCM3 outputs were used as off-line inputs to
the HyLand model, i.e. the models were not run
interactively.
The global pattern of change in temperature and
precipitation is similar in the seven scenarios, with the
temperature increase greatest at high latitude. Precipitation also increases at high latitudes and across much of
Asia. Precipitation decreases in Southern Africa, in the
Mediterranean, and in much of Latin America.
2.5. Land use data
Data on the historical expansion of cropland have
recently been published (Ramankutty and Foley, 1999).
These data cover the period 1700–1992 at approximately
30-year resolution, and encompass the global land
surface at 0.5 degree spatial resolution. These data were
averaged up to the 3.75 2.5 degree used by HadCM3
to prescribe the historical course of land use change in
the HyLand model. The data shows an increase in
cropland from around 2% of the global land area in
1700, to over 13% in 1990 (Fig. 2).
The SRES A1F, B1 and B2 scenarios include land use
data for the following categories: cropland, forest,
grassland, and ‘other’. No land use data exist for the
A2 scenario, so the A1F data were used instead. These
data are available for the four SRES regions at decadal
intervals between 1990 and 2100 (Fig. 2). As no data
with higher spatial resolution were available, the SRES
regional data were used to derive trends in land use
relative to 1990, which were applied to all cells within
that region. In this way, the appropriate temporal
change is represented whilst maintaining the more
complex spatial pattern at 3.75 2.5 degrees present in
1990. This also means that the discontinuity between the
historical and SRES data in Fig. 2 does not influence the
ARTICLE IN PRESS
P.E. Levy et al. / Global Environmental Change 14 (2004) 21–30
24
model. Fig. 2 shows that the area of cropland is
predicted to decline markedly in the A1 and B1
scenarios (after an initial rise in B1 until 2030). Only
in the B2 scenario is the area of cropland predicted to
increase.
At each point on the 3.75 2.5 degree grid, ten plots
were represented, each with a discrete land use type. As
the fraction of cropland increased over time, the
appropriate number of plots were switched from
the natural vegetation type, via the clear-felled state,
to the cropland type. The effect of land use type on the
modelled vegetation are described in Appendix A.
Where the fraction of cropland decreased over time,
the appropriate number of plots were switched from the
cropland type to the natural vegetation type. Thus both
cropland expansion and abandonment can be represented in the model.
2.6. Simulations
Details of the simulations performed are summarised
in Table 2. The HyLand model was run for a period of
400 years with pre-industrial conditions specified as
inputs: a CO2 concentration of 288 ppm, the HadCM3
climate for the decade 1860–1869, and the land use as
prescribed by Ramankutty and Foley (1999) for 1700.
During this period, the model state variables reached
equilibrium, and all values were saved at the end of this
spin-up period. Using these saved values for initialisation, the model was then run for a 290-year historical
period, from 1700 to 1990, using the historical CO2 data
from ISAM, the Ramankutty and Foley (1999) data for
land use change and the HadCM3 climate data for 1860
onwards. Again, all values were saved at the end of the
run and used as the initial values for the SRES scenario
runs.
Four simulations were run, corresponding to the A1F,
A2c, B1 and B2b scenarios. These covered the period
1990–2100 and used the SRES data for CO2, land use
change and the HadCM3 climate data. In a further set
of simulations for each SRES scenario, one set of these
inputs to the model was held constant at the 1990 level
over the period 1990–2100. Thus for each scenario, there
were three further simulations in which either: the CO2
concentration was held at 353 ppm; the 1990s climate
was used; or the 1990 land use pattern was retained. This
allowed the separation of the effects of climate change,
elevated CO2 and land use change. There thus 16
simulations in total for the period 1990–2100.
3. Results
Fig. 3 shows the global fluxes of carbon predicted by
HyLand between 1700 and 2100. The vegetation is
initially in equilibrium with the pre-industrial climate, so
net ecosystem productivity (NEP, i.e. net primary
productivity minus heterotrophic respiration) is near
zero before 1850. The flux resulting directly from land
use change (i.e. fire and log-removal, hereafter referred
to as the land use change flux, LUCF) is small during
this period, producing a small, negative net biome
productivity (NBP, equal to NEP plus LUCF). Fluxes
from the atmosphere to the land (sinks) are considered
to be positive.
Between 1850 and the present day, NEP gradually
becomes more positive, reaching 2.6 Pg C yr 1 in the
1990s. LUCF also increases markedly from around 1850
to the present day, and for most of this period, exceeds
NEP, resulting in NBP remaining negative until the
1970s. Though somewhat higher, the model predictions
for NBP in the 1980s and 1990s are close to the most
recent estimates based on inverse modelling of trends in
atmospheric CO2 and O2 concentrations (IPCC, 2001).
Similarly, the model predictions of LUCF are within the
limits of uncertainty on inventory or book-keeping-
Table 2
Scenarios for climate, CO2 (ppm) and land use (LU) used in four global-scale simulations
Simulation
A1F
‘Spin-up’ conditions
1700–1989
1990–2100
Climate
CO2
LU
Climate
CO2
LU
Preindustrial
(HadCM3
1860–1869)
288
Preindustriala
HadCM3
simulation
(from 1860
onwards)
288–352
CO2
LU
Ramankutty A1F–
and Foley
HadCM3
(1999)
353–970
A1F
A2c–
HadCM3
B1–
HadCM3
B2b–
HadCM3
353–856
A1F
353–549
B1
353–621
B2
A2
B1
B2
A1F, A2, B1 and B2 are IPCC SRES scenarios. HadCM3 is the UK Hadley Climate Model.
a
Based on the data of Ramankutty and Foley (1999) for 1700.
Climate
ARTICLE IN PRESS
P.E. Levy et al. / Global Environmental Change 14 (2004) 21–30
10
8
Net Ecosystem Productivity (NEP)
6
4
2
sink
0
source
-2
-4
8
Land Use Change Flux (LUCF)
a1f
a2c
b1
b2b
TAR
calYear vs NEP
a1f-calYear vs a1f-NEP
a2c-calYear vs a2c-NEP
b1-calYear vs b1-NEP
b2b-calYear vs b2b-NEP
Plot 1 Lower Control Line
sink Col 48 vs NEP_TAR
-1
Carbon flux (Pg C y )
10
6
4
2
0
source
-2
-4
10
8
Net Biome Productivity (NBP)
6
4
2
sink
0
-2
1700
source
1800
1900
2000
2100
Year
Fig. 3. Global fluxes of carbon predicted using the SRES A1F, A2c,
B1 & B2b scenarios for climate and CO2. Fluxes shown are net
ecosystem productivity (NEP, net primary productivity minus heterotrophic respiration), the flux resulting directly from land use change
(LUCF), and net biome productivity (NBP, NEP plus LUCF). Fluxes
from the atmosphere to the land (sinks) are denoted positive
throughout. Also shown are independent values based on inverse
modelling of trends in atmospheric CO2 and O2 concentrations from
the Third Assessment Report of the IPCC (TAR).
25
LUCF in the early part of the 21st century, whereafter it
returns to near zero, as in the A1F scenario. In both
these scenarios, the decline in cropland area (Fig. 2)
contributes strongly to the high NEP, as abandoned
cropland is converted back to the natural state where
regrowth of woody vegetation is permitted. Note that
this flux arises from land use change indirectly, so is
accounted for as NEP not LUCF under the definitions
given above. NBP results from the net effect of NEP and
LUCF, and as LUCF is generally small, it largely
follows the pattern in NEP. The terrestrial vegetation
thus acts as a net sink over the 21st century in all
scenarios, with the peak values ranging from 4 to
8 Pg C yr 1 in the different scenarios.
Fig. 4 shows the effects of climate change, elevated
CO2 and land use change on the simulated NBP over the
period 1990–2100. These are calculated as the difference
between NBP in the standard SRES scenario and that
when the given input is held constant. In all scenarios,
elevated CO2 has the largest influence on NBP, and this
effect is greatest in the A1F and A2C scenarios, where
CO2 concentrations reach the highest values. The effect
of climate change is considerably smaller, and acts to
reduce NBP in all scenarios. The influence of land use
change is of a similar magnitude, but may be positive or
negative, depending on the scenario (note that here,
both direct and indirect effects of land use change are
included in this measure). The B2B scenario is the only
one in which croplands increase, resulting in a net source
from deforestation. In the other scenarios, cropland is
abandoned, and natural vegetation permitted to increase, resulting in a net sink to forest regrowth.
Fig. 5 shows the global distribution of total ecosystem
carbon (plant plus soil) predicted by the model for 2100.
10
elevated CO2
climate change
land use change
All factors
-1
based estimates (also from IPCC, 2001). The residual of
the inverse modelling-based NBP and the inventorybased LUCF values gives the best available estimates of
NEP based (indirectly) on observations. The almost
exact correspondence with the model prediction of NEP
in the 1980s is probably spurious, as the uncertainty in
the observation is so large.
After 1990, there is a continuing rise in NEP following
the rise in CO2 concentration and temperature. This
lasts until around the middle of the century in most
scenarios, then begins to decline, to near zero in the B2
scenario. The exception is the A2 scenario, where NEP
continues to rise over the whole century, though more
slowly in the latter half. In the A1F scenario, LUCF
returns to near zero after 1990, as the scenario predicts
little further conversion of natural vegetation to cropland. In the B1 scenario, there is an initial peak in
Carbon flux (Pg C y )
8
6
4
2
sink
0
source
-2
-4
a1f
a2c
b1
b2b
SRES scenario
Fig. 4. Influence of the three main input factors on the predicted NBP
over the period 1990–2100 in four SRES scenarios (A1F, A2C, B1 &
B2B). Grey bars show the predicted value when all factors are included
in the simulation; other bars show the contribution from climate
change, elevated CO2 and land use change to this value. Note that
there are interaction effects.
ARTICLE IN PRESS
26
P.E. Levy et al. / Global Environmental Change 14 (2004) 21–30
The spatial pattern in all scenarios is roughly similar,
and broadly follows the current distribution of vegetation biomass. Agreement between model predictions for
the current day and the independent data of Olson et al.
(1983) suggest we can have a degree of confidence in the
model (Levy et al., in preparation). Differences between
the scenarios are more apparent if the change in
ecosystem carbon over the simulation is plotted rather
than the total (Fig. 6). This shows that most regions are
predicted to be carbon sinks, with the exceptions of
Amazonia, the Sahel, Central Australia, Central USA
(not in A2), and some areas of Eastern Europe, Russia
and China (B2 only). The strong source from the
Amazonian region is a consistent feature in all scenarios.
Fig. 7 shows the geographical distribution of the
effects of climate change, elevated CO2 and land use
change on the simulated NBP in the B2 scenario. This
shows that the predicted loss of carbon in Amazonia,
the Sahel and Central Australia is related to climate
change, as these sources are not seen when climate is
held constant (Fig. 7b). When CO2 is held constant,
large areas of the globe become net sources of carbon,
Fig. 5. Predicted total carbon (plant+soil, kg C m 2) in 2100 in the four SRES scenarios.
Fig. 6. Predicted change in total carbon (plant+soil, kg C m 2) between 1990 and 2100 in the four SRES scenarios.
ARTICLE IN PRESS
P.E. Levy et al. / Global Environmental Change 14 (2004) 21–30
including the African rainforest and large parts of the
South American forest and S.E. Asia (Fig. 7c). The
relatively small predicted source regions in Central Asia,
China and India are related to land use change, as these
sources are not seen in Fig. 7d.
Fig. 8 shows the predicted change in total ecosystem
carbon in the four SRES scenarios, classified by
vegetation type. The biggest carbon gains are found in
27
the broad-leaved forest class, and this has the highest
median value in all scenarios, though the largest losses
are also found in this class. Grasslands and needleleaved forest rank second or third depending on the
scenario. Needle-leaved forest show notably smaller
variation. The spatial distribution of vegetation types
varies little between scenarios, largely because the
spatial patterns in the input variables specified by the
Fig. 7. Predicted change in total carbon (plant+soil, kg C m 2) between 1990 and 2100 in the B2b SRES scenario where one set of these inputs to the
model was held constant at the 1990 level. In addition to the standard scenario (a), there were three further simulations in which either: the 1990s
climate was used (b); the CO2 concentration was held at 353 ppm (c); or the 1990 land use pattern was retained (d).
Change in total C, 1990 to 2100 (kg C m-2)
20
A1f
B1
A2
B2
10
0
-10
20
10
0
-10
Desert
Grassland
Bdlvd forest
vegetation type
Ndlvd forest
Desert
Grassland
Bdlvd forest
Ndlvd forest
vegetation type
Fig. 8. Boxplots showing the distribution of the losses and gains of total carbon (plant+soil, kg C m 2) between 1990 and 2100 in the four SRES
scenarios, classified by vegetation type. Boxes show the interquartile range with the horizontal line indicating the median. The vertical lines extend
from the box to the lowest and highest observations that are still inside the region defined by the following limits: Q1 1.5 (Q3 Q1) or Q3+1.5
(Q3 Q1), where Q1 and Q3 are the first and third quartiles. Outlying points outside of these limits are plotted as asterisks.
ARTICLE IN PRESS
28
P.E. Levy et al. / Global Environmental Change 14 (2004) 21–30
four SRES scenarios are generally similar. The dominant type of vegetation at any particular location is
therefore generally the same in the different scenarios,
irrespective of carbon gains or losses.
4. Discussion
The terrestrial biosphere is predicted to be a net sink
for carbon during the 21st century in all four SRES
scenarios simulated. As CO2 is the dominating influence
on the vegetation (enhancing net primary productivity)
the scenarios with high fossil fuel emissions, and thus the
highest CO2 concentrations (A1F & A2) have the largest
net sink for carbon. This would change if these scenarios
also assumed continued deforestation and cropland
expansion. The spatial pattern in vegetation losses and
gains is broadly similar in all scenarios, with losses
consistently being seen in Amazonia, the Sahel, South
Central USA and Central Australia. These are the
regions predicted to be at greatest risk from the
deleterious effects of the simulated global environmental
change. The spatial distribution of vegetation types is
very similar in all scenarios, and there are not major
shifts in the dominant types. Without the beneficial
effects of elevated CO2, the effects of climate change
(warming and drought) are much more severe. This is of
some serious concern, as the long-term and large-scale
effects of elevated CO2 are still open to question.
Under all SRES scenarios simulated, the terrestrial
biosphere is predicted to be a net sink for carbon over
practically all of the 21st century. This sink peaks
around 2050 and then diminishes rapidly towards the
end of the century as a result of climate change.
Averaged over the period 1990–2100, the net sink varies
between scenarios, from B2 to 6 Pg C yr 1. This sink is
the balance between the effects of CO2, climate and land
use change. Of these, CO2 is the dominant influence
producing the net sink. The effect of climate change can
be positive or negative, depending on the location, but
on balance has a negative effect. This largely results
from the decline in the Amazonian forest areas where
rainfall declines markedly. The reduction in rainfall in
the Amazon basin is not reproduced in all GCMs, so
there is considerable uncertainty in this result.
The effect of land use change can also be positive or
negative, depending on whether cropland is contracting
or expanding in the scenario (i.e. whether natural
vegetation is regrowing or being deforested). Here,
cropland contracts in two out of the three land use
scenarios. This may seem to be a rather optimistic view
of future world development, in that global forest area
does not decline overall, despite expected continued
tropical deforestation, and cropland areas do not
generally expand despite continued population growth.
There is large uncertainty in the future trend in cropland
area as population increases, as illustrated in Fig. 9.
Until very recently, cropland area has shown a linear
increase with population. However, in the last 50 years,
in which the population has doubled, the increase in
cropland area has been very small. How this trend will
continue with a possible further doubling of the
population over the next century is clearly highly
uncertain. The three SRES scenarios in Fig. 9 cover
the possibilities of a slight increase in cropland (B2), and
of a substantial decrease (A1F and B1). We suggest that
the possibility of substantial increase should also be
included. Recent studies have estimated that, based on
soil, water and climatic constraints, the area of cropland
could still be increased by an additional 120% of the
current value, i.e. to nearly 30% of the land area (Bot
et al., 2000; Ramankutty et al., 2002). This expansion
would necessarily result in substantial deforestation.
Deforestation of areas for permanent pasture rather
than cropland may be a significant term, particularly in
South America. However, this land use change is not
included in these simulations as (i) the SRES land use
data do not distinguish natural grassland from that
which is artificially created, and (ii) there are no data
available on the historical expansion of pasture land.
Appendix A. Model description
The Hybrid model, fully described elsewhere (Friend
et al., 1997; Friend and White, 2000) was simplified,
taking a more top–down approach, removing the
individual-tree simulation in the original ‘gap’ model,
retaining all the vegetation and soil compartments, but
with simplified hydrology and N dynamics. The model
uses a daily time-step. Inputs were climate variables,
atmospheric CO2 concentration, and land use change.
Here we describe the major differences from Friend et al.
(1997) and note where the changes described in Friend
and White (2000) are included.
A.1. Vegetation
*
*
Three (cf. eight) plant types were represented: herbs,
broad-leaved and need-leaved trees. Individual trees
were not represented. The vertical dimension in the
canopy was ignored, except that the herbaceous layer
was implicitly beneath the tree canopy.
Plant respiration (i.e. respiration not accounted for in
net daytime photosynthesis) was assumed to be half
of gross photosynthesis. Several recent studies have
indicated that this assumption may be a good
approximation of reality (Gifford, 1995; Waring
et al., 1998; Dewar et al., 1999; Thornley and
Cannell, 2000), better than the assumption that maintenance respiration is a linear function of the total
carbon content of each of the plant compartments
ARTICLE IN PRESS
P.E. Levy et al. / Global Environmental Change 14 (2004) 21–30
A1F
29
B2
B1
15
2100
1990
1970
1990
Fcrop (%)
1950
1930
10
2000
1910
1900
2020 2030
2050
2010
2100
2060
2080
2090
2080
1850
1800
1750
1700
5
2100
0
0
1000
2000
3000
4000
5000
5000
Population (millions)
6000
7000
8000
9000
10000
11000
Population (millions)
Fig. 9. Relationship between global population and the area of cropland as a fraction of the global land area (Fcrop,%), from historical sources (lefthand panel) and SRES scenarios (right-hand panel).
*
*
*
*
*
and an exponential function of air temperature
(Ryan, 1991).
Nitrogen dynamics were not simulated, but the
nitrogen content of the top layer of foliage (Ntop)
was prescribed as 1.3 kg N m 2.
NPP was allocated in fixed ratios to foliage (0.1 for
trees, 0.28 for herbs) structure (0.7 for trees, 0.33 for
herbs) and roots (0.2 for trees, 0.39 for herbs).
Leaf area index was the product of foliage carbon
and specific leaf area (36 m2 kg C 1 for herbs and
broad-leaved trees, 18 m2 kg C 1 for needle-leaved
trees).
Litter fall was calculated from fixed turnover rates for
foliage (0.8, 1 and 6 years for herbs, broad-leaved and
needle- leaved trees, respectively) structure (0.8, 20,
10 years for herbs, broad-leaved and needle-leaved
trees, respectively) and roots (0.5 years for all types).
If the stem carbon of one tree type was within 750%
of that of the other, then PAR was shared equally; if
not the lesser type was excluded. Herbs dominated if
the woody canopy was sparse. PAR at the top of the
canopy was scaled to ensure that the sum of light
used did not exceed that available. All plant types
were initialised at 0.1 kg C m 2 allocated between
structural compartments with the same fractions as
for NPP (see above).
A.2. Soil
*
*
Soil water capacity was calculated as a linear function
of soil carbon (Friend and White, 2000). When the
soil contained 10% more than its capacity there was
drainage. Soil water-filled space was calculated from
soil water content and a prescribed soil texture as
described by Friend and White (2000). Soil water
potential was calculated from soil water content.
In order to deal effectively with deforestation, pools
of coarse surface and below-ground structural litter
*
were added to the decomposition sub-model of
Friend et al. (1997) with residence times of 73 and
53 years, respectively, when soil temperature and
moisture factors were unity. These long residence
times delayed carbon loss after clearfelling (see
below).
Since soil carbon may take centuries to reach
equilibrium, the pool size was initialised in the
‘spin-up’ run with an estimated soil carbon content
derived from the observed relationship between soil
carbon and pre-industrial precipitation and temperature (Friend and White, 2000).
A.3. Land use
Three land use options were employed, which were
assumed to alter carbon pools and fluxes as follows:
*
*
Natural vegetation: No constraints placed upon the
simulated vegetation.
Deforestation (clearfelled forest): (i) It was assumed
that 64% of the above-ground stem carbon was
removed instantly by fire or as logs. No account was
taken of the fate of forest products. Houghton (1991)
assumed that 33% of the original vegetation was left
to decay on site at the time of clearing; thus 67% was
lost to burning or removal. If 30% of stem mass is
stump and below-ground and all fine plus 30% of
coarse above-ground litter is burned (Hao et al.,
1990), then about 64% of above-ground stem must be
removed to leave 33% of the original carbon on site
(70% of stem is assumed to be coarse). (ii) Clearcutting was immediately followed by a fire, which
oxidized 30% of coarse above-ground stem litter, all
other litter, and all above-ground herbaceous plant
parts (Hao et al., 1990). Charcoal production was
ignored. (iii) The remaining litter was apportioned to
coarse and fine litter above- and below-ground. (iv)
Soil disturbance in the year of deforestation was
ARTICLE IN PRESS
P.E. Levy et al. / Global Environmental Change 14 (2004) 21–30
30
*
assumed to cause 30% of the carbon in protected
pools to move to the active decomposable pools.
Cropland. (i) Tree regeneration was prevented. (ii)
Cultivation caused 30% of the carbon in protected
pools to move to the active decomposable pools every
year. (iii) Harvesting removed 50% of above-ground
vegetation carbon every year. No account was taken
of the fate of crop products. The remaining carbon
was transferred to litter. (iv) Incorporation of litter
was simulated by assuming that 50% of the aboveground structural and metabolic litter pools were
transferred to the topsoil structural and metabolic
litter pools each year of cultivation (Voroney and
Angers, 1995; van Veen and Kuikman, 1990).
References
Bot, A.J., Nachtergaele, F.O., Young, A., 2000. Land Resource
Potential and Constraints at Regional And Country Levels. FAO,
Rome, p. 122.
Comins, H.N., McMurtrie, R.E., 1993. Long-term response of
nutrient-limited forests to CO2 enrichment-equilibrium behavior
of plant-soil models. Ecological Applications 3, 666–681.
Cox, P.M., Betts, R.A., Jones, C.D., Spall, S.A., Totterdell, I.J., 2000.
Acceleration of global warming due to carbon-cycle feedbacks in a
coupled climate model. Nature 408, 184–187.
Cramer, W., Kicklighter, D.W., Bondeau, A., Moore, B., Churkina,
C., Nemry, B., Ruimy, A., Schloss, A.L., 1999. Comparing global
models of terrestrial net primary productivity (NPP): overview and
key results. Global Change Biology 5, 1–15.
DeFries, R.S., Field, C.B., Fung, I., Collatz, G.J., Bounoua, L., 1999.
Combining satellite data and biogeochemical models to estimate
global effects of human-induced land cover change on carbon
emissions and primary productivity. Global Biogeochemical Cycles
13, 803–815.
Dewar, R.C., Medlyn, B.E., McMurtrie, R.E., 1999. Acclimation of
the respiration photosynthesis ratio to temperature: insights from a
model. Global Change Biology 5, 615–622.
Farquhar, G.D., von Caemmerer, S., 1982. Modelling of photosynthetic response to environmental conditions. In: Lange, O.L.,
Nobel, P., Osmond, C.B., Ziegler, H. (Eds.), Physiological Plant
Ecology II. Water Relations and Carbon Assimilation, Vol. 12B.
Springer, Berlin, pp. 549–587.
Friend, A.D., 1995. Pgen—an integrated model of leaf photosynthesis,
transpiration, and conductance. Ecological Modelling 77, 233–255.
Friend, A.D., White, A., 2000. Evaluation and analysis of a dynamic
terrestrial ecosystem model under preindustrial conditions at the
global scale. Global Biogeochemical Cycles 14, 1173–1190.
Friend, A.D., Stevens, A.K., Knox, R.G., Cannell, M.G.R., 1997. A
process-based, terrestrial biosphere model of ecosystem dynamics
(Hybrid v3.0). Ecological Modelling 95, 249–287.
Gifford, R.M., 1995. Whole plant respiration and photosynthesis of
wheat under increased CO2 concentration and temperature: longterm vs short-term distinctions for modelling. Global Change
Biology 1, 385–396.
Hao, W.M., Liu, M.-H., Crutzen, P.J., 1990. Estimates of annual and
regional releases of CO2 and other trace gases to the atmosphere
from fires in the tropics, based on the FAO statistics for the period
1975–80. In: Goldammer, J.G. (Ed.), Fire in the Tropical Biota.
Ecosystem Processes and Global Challenges, Vol. 84. Springer,
Berlin, pp. 440–462.
Houghton, R.A., 1991. Releases of Carbon to the atmosphere from
degradation of forests in Tropical Asia. Canadian Journal of
Forest Research–Revue Canadienne De Recherche Forestiere 21,
132–142.
Houghton, R.A., 1995. Land-use change and the carbon-cycle. Global
Change Biology 1, 275–287.
IPCC, 2000. Emissions Scenarios. A Special Report of Working Group
II of the Intergovernmental Panel on Climate Change. In:
Cambridge University Press, Cambridge 612pp.
IPCC, 2001. Climate Change 2001: the scientific basis. Summary for
Policymakers, IPCC, Geneva, p. 881.
Jarvis, P.G., 1976. The interpretation of the variations in leaf water
potential and stomatal conductance found in canopies in the field.
Philosophical Transactions of the Royal Society of London (B)
273, 593–610.
Jarvis, P.G., 1998. European Forests and Global Change: the likely
Impacts of Rising CO2 and Temperature, Cambridge University
Press, Cambridge. p. 380.
McGuire, A.D., Sitch, S., Clein, J.S., Dargaville, R., Esser, G., Foley,
J., Heimann, M., Joos, F., Kaplan, J., Kicklighter, D.W., Meier,
R.A., Melillo, J.M., Moore, B., Prentice, I.C., Ramankutty, N.,
Reichenau, T., Schloss, A., Tian, H., Williams, L.J., Wittenberg,
U., 2001. Carbon balance of the terrestrial biosphere in the
twentieth century: analyses of CO2, climate and land use effects
with four process-based ecosystem models. Global Biogeochemical
Cycles 15, 183–206.
Monteith, J.L., Unsworth, M.H., 1990. Principles of Environmental
Physics. Edward Arnold, London, p. 291.
Olson, J.S., Watts, J.A., Allison, L.J., 1983. Carbon in Live Vegetation
of Major World Ecosystems. Oak Ridge National Laboratory, Oak
Ridge, Tennessee, p. 154.
Ramankutty, N., Foley, J.A., 1999. Estimating historical changes in
global land cover: croplands from 1700 to 1992. Global Biogeochemical Cycles 13, 997–1027.
Ramankutty, N., Foley, J.A., Norman, J., McSweeney, K., 2002. The
global distribution of cultivable lands: current patterns and
sensitivity to possible climate change. Global Ecology and
Biogeography 11, 377–392.
Ryan, M.G., 1991. Effects of climate change on plant respiration.
Ecological Applications 1, 157–167.
Sellers, P.J., Berry, J.A., Collatz, G.J., Field, C.B., Hall, F.G., 1992.
Canopy reflectance, photosynthesis, and transpiration 0.3. A
reanalysis using improved leaf models and a new canopy
integration scheme. Remote Sensing of Environment 42, 187–3216.
Stewart, J.B., 1988. Modelling surface conductance of pine forest.
Agricultural and Forest Meteorology 43, 19–335.
Thornley, J.H.M., Cannell, M.G.R., 2000. Modelling the components
of respiration: representation and realism. Annals of Botany 85,
55–67.
van Veen, J.A., Kuikman, P.J., 1990. Soil structure. Biogeochemistry
11, 213–233.
Voroney, R.P., Angers, D.A., 1995. Analysis of the short-term effects
of management on soil organic matter using the Century model. In:
Lal, R., Kimble, J., Levine, E., Stewart, B.A. (Eds.), Soil
Management and Greenhouse Effect. Lewis Publishers, Boca
Raton, pp. 113–120.
Waring, R.H., Landsberg, J.J., Williams, M., 1998. Net primary
production of forests: a constant fraction of gross primary
production? Tree Physiology 18, 129–134.